Parallel local search in comet

The availability of commodity multiprocessors offers significant opportunities for addressing the increasing computational requirements of optimization applications. To leverage these potential benefits, it is important however to make parallel processing easily accessible to a wide audience of optimization programmers. This paper addresses this challenge by proposing parallel programming abstractions that keep the distance between sequential and parallel local search algorithms as small as possible. The abstractions, that include parallel loops, interruptions, and thread pools, are compositional and cleanly separate the optimization program and the parallel instructions. They have been evaluated experimentally on a variety of applications, including facility location and coloring, for which they provide significant speedups.